1. Field of the Invention
This invention relates to the monitoring of electro-physiological signals of a subject in a harsh electrical environment. This invention further relates to the monitoring of EEG signals of a patient, while said patient is being operated on with an electrical surgical knife.
2. Brief Description
Disclosed, herein, is a method and apparatus that significantly limits the effect of high frequency (“HF”) interferences on acquired electro-physiological signals, such as the EEG and EMG. This method comprises of two separate electronic circuitries and processing means. One circuitry is used to block the transmission of HF interferences to the instrumentation amplifiers. It is comprised of a front-end active filter, a low frequency electromagnetic interference (“EMI”) shield, and an isolation barrier interface which isolates the patient from earth ground. The second circuitry is used to measure the difference in potential between the two isolated sides of the isolation barrier. This so-called “cross-barrier” voltage is directly representative of the interference level that the instrumentation amplifier is subjected to. This circuitry is used to confirm that the acquired signals are not corrupted by the interference. The processing means further use both the acquired electro-physiological signals and the cross-barrier voltage measurement to qualitatively or quantitatively assess the state or well-being of the patient.
3. Technical Review
Clinical environments can be particularly hostile for systems dedicated to real-time acquisition of electro-physiological signals. Surgical equipment, such as electro-surgical units (“ESU”), are a typical source of high frequency (HF) interferences (>100 kHz). Electrodes used to measure bio-potentials can be subject to common mode noise whose amplitude can reach several orders of magnitudes higher than the range the bio-signals such as ECG, EOG, EEG, EMG and the like, which need to be acquired. A slight difference between the noise at the recording electrode site and the noise at the reference electrode site usually induces the saturation of the instrumentation amplifier. In this case, the acquired signal cannot be salvaged, as it does not contain any viable information.
Critical applications that rely on electro-physiological data acquisition to continuously monitor the patient's state will therefore suffer from HF interferences. If the source signal is lost due to these interferences, these monitoring systems will not be able to output reliable information pertinent to the patient state. In systems where this information is used to provide or adjust a treatment, the loss of the source signal is cause for concern.
In the past, a number of researchers have disclosed methods, which attempt to alleviate HF interference in high gain instrumentation amplifiers. For example, in U.S. Pat. No. 4,537,200 to Windrow the method involves both passive and active input HF filters followed by proper isolation, and a hard-wired adaptive filter to remove the interference based on a reference electrode that provides a regressive channel used by the adaptive filter. U.S. Pat. No. 6,430,437 to Marro uses a combination of a passive and active input filters, a low isolation interface to minimize the leakage capacitance, and a low frequency shielding. U.S. Pat. No. 6,985,833 to Shambroom proposes the use of a hard-wired quantal (i.e., on/off) detection of the presence of ESU. When ESU is detected, a logic flag is raised. The computing means then rejects the current epoch and replaces it with the previously good epoch.
These three references are attempts to develop biopotential data acquisition systems that are capable of working in a harsh HF electrical environment. Unfortunately none are capable of actually measuring residue interferences that get through the front end filters. In the current state of the art this is absolutely critical if one requires real time monitoring. While Shambroom has proposed a circuitry that can detect the presence of ESU, then reject the corrupted epochs and finally replace the rejected epochs with previous epochs. This does not allow for the actual measuring of biopotentials while an ESU is in use.
Conversely Widrow and Marro have taken a different approach, and have attempted to filter out all interferences. This works well as long as the interferences are completely filtered out. Unfortunately, often in practice the use of an ESU will cause interference that is too great to be completely eliminated from the signal. In these cases, the acquired biosignals are still be corrupted, albeit with a corrupting noise that is now in the same order of magnitude as the desired signal itself. Therefore, it becomes very difficult to determine that the acquired biosignals are corrupted, and thus cannot be used. Thus the practicality of these systems is questionable in harsh RF environment.
In the presence of HF interferences, the previously mentioned technology either leaves the user with the explicit knowledge that the signals are corrupted, or leaves the user with signals whose integrity cannot be guaranteed. Therefore a need exists for a biosignal data acquisition system that can both filter out large HF interference, and measure the residual interferences to eventually correct them using digital signal processing means. This allows for accurate and reliable data to be acquired while in a harsh HF environment. It is an object of the present invention to provide a robust system that meets such a need.
The present invention differs in that a combination of active filters, isolation, shield, measurement circuitry, and software are used to deal with HF interferences. The present invention uses a first line of defense that is optimized using front-end hardware, which dramatically reduces the magnitude of the interference, as seen by the instrumentation amplifiers. The front-end hardware includes: an optimized active front-end HF filter that buffers the instrumentation amplifiers against HF interferences, a high isolation-barrier interface between the patient-side electronics and the computer-side electronics, and a low voltage shield.
As compared to Marro'437 and Widrow'200, this circuitry uses only one active filter at the front-end. The choice of an active filter vs. the passive HF filter architecture proposed by Marro'437 and Widrow'200 is dictated by the need for a high input impedance (>100 MΩ) in the instrumentation amplifier bandwidth, a requirement set forth by the International Federation of Clinical Neurophysiology (IFCN), as a pre-requisite for EEG and EMG data acquisition.
The role of the isolation barrier is two-fold: to provide the necessary patient/earth isolation required by the IEC 60601 standard for medical device safety, and to provide a difficult return-to-earth path for the HF interferences. Marro'437 attempted to achieved this by using opto-isolators to transmit the acquired data from the patient-side electronics to the computer-side electronics. Marro'437 mentions that only opto-isolators are suitable for this application, as they provide an ultra-low coupling capacitance between the patient-side and computer-side ground planes. The Applicants, however, found out unexpectedly that individual drum inductor coils loosely coupled end-to-end across the isolation barrier implements a pulse transformer to provide galvanic isolation can also be used effectively.
In many cases, the optimized front-end will block almost all of the HF interference. In many cases, the remaining HF noise that still perturbs the input of the instrumentation amplifier, is significantly less than the signal of interest, in which case the acquired data will only be marginally corrupted by the interference. In other cases, however, the HF interferences are so large or close to the recording electrode sites that the residual interference noise level is about the same order of magnitude as the signal of interest. The acquired signal is therefore corrupted by a wide-band noise, which will perturb any subsequent analysis and patient state determination. It is therefore necessary to properly detect such situations in order to take appropriate action.
Detecting wide-band noise in acquired data can be particularly difficult when the signals of interest are already wide-band noise-like signals, such as EEG and EMG signals. Systems equipped with only the front-end electronics, as described above, are practical only when HF interference are always completely blocked. Since this is not the case in practice, the present invention preferably uses the addition of a secondary circuitry whose role is to measure the difference in potential between the isolated patient-side electronics and the computer-side electronics. This circuitry provides a cross-barrier voltage measurement that represents the magnitude of the corrupting interference. In electrically quiet environments (no HF interference), or when the HF interference are small enough to be significantly attenuated by the first circuitry, this voltage will be low, close to 0. If the HF interference is large enough that it cannot be completely rejected, the measured cross-barrier voltage will be higher, thereby indicating that the acquired data may be corrupted.
In critical applications, such as patient monitoring, the present invention preferably uses a post-processing algorithm may opt for rejecting the acquired samples for which the cross-barrier voltage was high. In other embodiments, special techniques may be used to extract the signal of interest from the acquired data see for instance A wavelet based de-noising technique for ocular artifact correction of the electroencephalogram. Zikov, et al. Eng. in Med. and Bio. 24th Ann. Conf. p 98-105 vol. 1 (2002).
A device detecting the presence of HF noise was proposed by Shambroom'833. However, the described circuitry only documents the presence or absence of RF noise. In contrast, the present invention preferably measures the interference which corrupts the instrumentation amplifier. This information can be used dynamically by the processing means to assess the best course of action. Shambroom'833 proposes that corrupted samples be replaced by non-corrupted samples acquired previously. In contrast, the current invention proposes to filter the acquired EEG if the cross-barrier voltage is low enough, and reject completely the corrupted samples if the cross-barrier voltage is too large. The thresholds used to determine whether a sample must be kept, filtered, or rejected can be dynamically changed depending on the sensitivity to HF noise of the processing algorithms.